CN106868440A - A kind of strip continuous hot galvanizing thickness of coating prediction and its adjusting method - Google Patents

Abstract

The invention discloses a kind of prediction of strip continuous hot galvanizing thickness of coating and its adjusting method, thickness of coating forecast model is set up using regression analysis, thickness of coating predicted value is calculated in real time and air knife regulated quantity is calculated according to strip hot-dip galvanizing process data and current thickness of coating desired value.Thickness of coating Forecasting Methodology of the invention can real-time estimate thickness of coating value, and calculate air knife regulated quantity reference value, adjust air knife parameter in time in order to operative employee.Realize thickness of coating dynamic prediction, it is ensured that thickness of coating is rapidly achieved target thickness, it is to avoid thickness of coating fluctuates for a long time.

Description

A kind of strip continuous hot galvanizing thickness of coating prediction and its adjusting method

Technical field

The invention belongs to strip continuous hot galvanizing technical field, it is adaptable to strip Continuous Hot Dip Galvanizing Line thickness of coating Prediction and Operating Guideline.

Background technology

Strip hot-dip galvanizing process is a multi-variable system for complexity, and thickness of coating measurement has error or big Lag issues, cause zinc coat thickness control precision universal not high.Thickness of coating detection generally have hot calibrator and Cold state thickness gauge two ways, hot calibrator is arranged on air knife top, and detection is almost without delayed, but coating There is larger error in thickness measure.Strip hot-dip galvanizing production line uses cold state thickness gauge detection mode, cold conditions mostly Calibrator measures more accurate to thickness of coating, but generally apart from more than 100 meters of air knife, thickness of coating measurement In the presence of serious delayed.

Patent 1 (CN 101429639) strip hot-dip galvanizing line flash plating production method, it is true according to different steel specification The span of related process parameters is determined, there is certain directive function to production process.But actual production Middle hot galvanizing process is multivariable influence, and technological parameter can only set as the initial of zinc coat thickness control in patent Definite value, it is impossible to meet the requirement of thickness of coating dynamic regulation.And strip speed would generally be according to the operation in stove area Ability changes, and this is accomplished by operating personnel and adjusts air knife parameter in time according to strip speed change, eliminates plating Layer thickness fluctuations.

A kind of device for measuring zinc coating thickness in Continuous Hot Dip Galvanizing of patent 2 (CN 102465246) And method, the zinc coating thickness that invention is capable of achieving on strip ad-hoc location is accurate corresponding with air knife regulated quantity, can For operative employee provides Operating Guideline, thickness of coating is in measurement of coating thickness with the corresponding result of air knife regulated quantity in invention What instrument measurement was calculated after terminating, and after calibrator is typically mounted on 100 meters of air knife, thickness of coating and air knife The corresponding result of regulated quantity exist it is serious delayed, cause the thickness of coating of operating personnel adjust exist it is delayed.

The content of the invention

It is an object of the invention to a kind of strip continuous hot galvanizing thickness of coating Forecasting Methodology for proposing, can be real-time Prediction thickness of coating value, and air knife regulated quantity reference value is calculated, adjust air knife parameter in time in order to operative employee. Realize thickness of coating dynamic prediction, it is ensured that thickness of coating is rapidly achieved target thickness, it is to avoid when thickness of coating is long Between fluctuate.

1 present invention sets up thickness of coating forecast model using regression analysis, and strip heat is read by online Galvanizing process data, real-time estimate calculates current thickness of coating value, and according to current thickness of coating setting value meter Air knife regulated quantity is calculated, for operating personnel provide Operating Guideline.Mode input variable have air knife blast, air knife with The distance of strip, strip speed, model are output as thickness of coating；Invention sets up coating using regression analysis Thickness prediction model, model parameter estimation is carried out using least square method.

1) Sample Data Collection.According to thickness of coating 80g/m²、100g/m²、120g/m²、180g/m²、 276g/m²Five kinds of specifications collect sample data.Each specification 40 sample datas of collection, collection different steel, Air knife blast actual value under the conditions of distance of friction speed, different air knife blast, different air knives and strip etc., Air knife and strip apart from actual value, strip speed actual value, the actual Value Data such as thickness of coating.

2) multiple linear regression modeling.

Analysis is carried out back using power function model：

CW=kV^aD^bP^c (1)

P- air knife blast actual values；D- air knives are with strip apart from actual value；V- strip speed actual values； CW- thickness of coating actual values；Model parameter k, a, b, c to be estimated.

Linearization process is carried out to formula (1)：

LnCW=Lnk+aLnV+bLnD+cLnP

CW*=LnCW, k*=Lnk, V*=LnV, D*=LnD, P*=LnP, above formula is made to be expressed as：

CW*=k*+aV*+bD*+cP*

Sample data is pre-processed.Sample data P, D of collection, V, CW are carried out into logarithm operation, calculate P*, The process datas such as D*, V*, CW*.

Multiple linear regression modeling is carried out using process datas such as P*, D*, V*, CW*, using EXCEL Or the DAS such as SPSS carries out multiple linear regression analysis, using Least Square Method model parameter K, a, b, c value, finally set up thickness of coating forecast model, and real-time estimate is carried out to thickness of coating.

P- air knife blast actual values；D- air knives are with strip apart from actual value；V- strip speed actual values； CW₈₀~CW₂₇₆The thickness of coating predicted value of-different size；k₁~k₅、a₁~a₅、b₁~b₅、c₁~c₅Estimate Model parameter.

2 result of calculations for showing forecast model in real time in thickness of coating operation man-machine interface, if the plating for calculating Layer thickness value shows the alarm signal of operating mistake below or above the 10% of thickness of coating desired value in man-machine interface Breath, points out operative employee to be intervened manually in time.

3 calculate with reference to regulated quantity

The regulated quantity such as air knife blast or air knife and strip distance is calculated in real time using coating forecast model, is operator Member provides operation reference.When air knife blast is less than 450mbar, using air knife blast shaping modes, air knife is calculated Blast regulated quantity；When air knife blast is equal to or higher than 450mbar, using air knife and strip apart from shaping modes, Air knife is calculated with strip apart from regulated quantity.Operative employee can be operated manually according to the regulated quantity.

P_set_ air pressure regulated quantity reference value, D_set- air knife spacing regulated quantity reference value, W_set- thickness of coating mesh Scale value.

The present invention has following characteristics and beneficial effect：

Strip Continuous Hot Dip Galvanizing Line thickness of coating prediction proposed by the present invention and its adjusting method, can be dynamic Current strip coating thickness is predicted, air knife parameter is adjusted in time in order to operative employee, so as to ensure that thickness of coating is fast Speed reaches target set point.

Brief description of the drawings

Fig. 1-80g/m²The thickness of coating prediction curve of specification；

Fig. 2-100g/m²The thickness of coating prediction curve of specification；

Fig. 3-120g/m²The thickness of coating prediction curve of specification；

Fig. 4-180g/m²The thickness of coating prediction curve of specification；

Fig. 5-276g/m²The thickness of coating prediction curve of specification.

Specific embodiment

1 by taking certain steel mill's galvanization production line as an example, sets up thickness of coating forecast model.Different size is pressed first Thickness of coating collects sample data, and carries out linearization process.

180g/m²Specification

100g/m²Specification

120g/m²Specification

180g/m²Specification

276g/m²Specification

2 multiple linear regressions are modeled.Application data analysis software EXCE softwares carry out multiple linear regression point Analysis, uses Least Square Method model parameter, sets up thickness of coating forecast model.

Target plating thickness 80g/m²：CW₈₀=0.731V^1.157D^1.246P^-0.885

Target plating thickness 100g/m²：CW₁₀₀=1.777V^0.979D^0.942P^-0.801

Target plating thickness 120g/m²：CW₁₂₀=1.619V^0.985D^0.883P^-0.753

Target plating thickness 180g/m²：CW₁₈₀=2.771V^0.667D^0.707P^-0.594

Target plating thickness 276g/m²：CW₂₇₆=V^1.073D^1.101P^-0.593

Using historical data inspection model prediction effect, as Figure 1-5, thickness of coating predicted value and plating thickness Deviation is less than 10% between degree actual value, can carry out commercial Application.

Produce reality data are sent into model by 3, and thickness of coating value is calculated in real time.Operated in thickness of coating man-machine The result of calculation of forecast model is shown on interface in real time, if the thickness of coating value for calculating is below or above plating thickness The 10% of degree desired value, the warning message of operating mistake is shown in man-machine interface, points out operative employee to carry out hand in time It is dynamic to intervene.

4 regulated quantitys are calculated, and air knife blast is adjusted less than 450mbar using air knife blast, is with reference to regulated quantity：

Air knife blast is equal to or higher than 450mbar and is adjusted with the distance of strip using air knife, is with reference to regulated quantity：

The reference regulated quantity of other specification thickness of coating calculates identical with this.

Claims (3)

1. a kind of strip continuous hot galvanizing thickness of coating is predicted and its adjusting method, it is characterised in that using recurrence Analysis method sets up thickness of coating forecast model, and thickness of coating value is calculated in real time and number of passes is crossed according to strip hot-dip galvanizing Air knife regulated quantity is calculated according to current thickness of coating desired value, is comprised the following steps：

1) Sample Data Collection, collects sample data, and carry out linearization process by different size thickness of coating； The sample data includes the distance of different steel, friction speed, different air knife blast, different air knives and strip Under the conditions of air knife blast actual value, air knife and strip apart from actual value, strip speed actual value, plate thickness The actual Value Data of degree；The linearization process that carries out is that the sample data of collection is carried out into logarithm operation；

2) multiple linear regression modeling

Analysis is carried out back using power function：

CW=kV^aD^bP^c (1)

P_-Air knife blast actual value；D_-Air knife is with strip apart from actual value；V_-Strip speed actual value； CW_-Thickness of coating actual value；Model parameter k, a, b, c to be estimated；

Linearization process is carried out to formula (1)：

LnCW=Lnk+aLnV+bLnD+cLnP

Make CW^*=LnCW, k^*=Lnk, V^*=LnV, D^*=LnD, P^*=LnP, above formula is expressed as：

CW^*=k^*+aV^*+bD^*+cP^*；

By the data of linearization process, carry out multiple linear using EXCEL or SPSS DAS and return Return analysis, using Least Square Method model parameter k, a, b, c value, finally set up thickness of coating prediction Model, real-time estimate is carried out to thickness of coating；

3) produce reality data are sent into model, thickness of coating value is calculated in real time, shown in real time in man-machine interface Show the result of calculation of forecast model, if the thickness of coating value for calculating is below or above thickness of coating desired value 10%, show the warning message of operating mistake in man-machine interface；

4) regulated quantity is calculated

Air knife blast or air knife are calculated in real time with strip apart from regulated quantity using coating forecast model, and air knife blast is low When 450mbar, using air knife blast shaping modes, air knife blast regulated quantity is calculated；Air knife blast be equal to or During higher than 450mbar, using air knife with strip apart from shaping modes, air knife is calculated with strip apart from regulated quantity；

$\left\{\begin{array}{cc}{P}_{set}=\frac{{{\mathrm{CW}}_{set}}^{\frac{1}{c}}}{k^{\frac{1}{c}}{V}^{\frac{a}{c}}{D}^{\frac{b}{c}}}& P_{set}<450mbar\\ D_{set}=\frac{{{\mathrm{CW}}_{set}}^{\frac{1}{b}}}{k^{\frac{1}{b}}{V}^{\frac{a}{b}}{P}^{\frac{c}{b}}}& P_{set}\&GreaterEqual;450mbar\end{array}\right.$

P_{set_}Air pressure regulated quantity reference value, D_{set_}Air knife spacing regulated quantity reference value, W_{set_}Thickness of coating mesh Scale value.

2. a kind of strip continuous hot galvanizing thickness of coating prediction according to claim 1 and its adjusting method, It is characterized in that：The different size thickness of coating includes thickness of coating 80g/m²、100g/m²、120g/m²、 180g/m²、276g/m²Five kinds of specifications.

3. a kind of strip continuous hot galvanizing thickness of coating prediction according to claim 2 and its adjusting method, It is characterized in that the thickness of coating forecast model corresponding to different size thickness of coating is：

${\mathrm{CW}}_{80}=k_1{V}^{a_1}{D}^{b_1}{P}^{c_1}$

${\mathrm{CW}}_{100}=k_2{V}^{a_2}{D}^{b_2}{P}^{c_2}$

${\mathrm{CW}}_{120}=k_3{V}^{a_3}{D}^{b_3}{P}^{c_3}$

${\mathrm{CW}}_{180}=k_4{V}^{a_4}{D}^{b_4}{P}^{c_4}$

${\mathrm{CW}}_{276}=k_5{V}^{a_5}{D}^{b_5}{P}^{c_5}$

Wherein：CW_80~CW_{276_}The thickness of coating predicted value of different size；k₁~k₅、a₁~a₅、b₁~b₅、 c₁~c₅The model parameter estimated.